Dependence of Phospholipase D1 Multi-monoubiquitination on Its Enzymatic Activity and Palmitoylation*

Phospholipase D (PLD) is an important lipase in many cellular processes, including vesicular trafficking, cell survival, and cell migration. In the present study, we show that PLD1, but not PLD2, is posttranslationally modified by multi-monoubiquitination. Intriguingly, suppression of lipase activity either by mutation of the HKD motif (PLD1 H896R, K898R, or D903A) or the phosphatidylinositol 4,5-bisphosphate binding motif (PLD1 R691G,R695G) or through use of PLD-selective inhibitors impaired the ubiquitination of PLD1, although stimulation of lipase activity by phorbol 12-myristate 13-acetate did not enhance its ubiquitination. A palmitoylation-deficient mutant PLD1 allele, which exhibits altered patterns of vesicular trafficking, had significantly lower levels of monoubiquitination. In addition, the expression of ubiquitin-fused PLD1 induced aberrantly enlarged vesicles partially co-localized with the Golgi complex but not with early endosomes. The altered localization was reduced by the K898R mutation, suggesting a role of multi-monoubiquitination in PLD1 subcellular localization. Surprisingly, the degradation of PLD1, but not of PLD1 K898R or PLD2, was blocked by inhibitors of proteasomes but not by inhibitors of lysosomes or other proteases, suggesting a role of the ubiquitination in proteasomal degradation of PLD1. In summary, our studies show that PLD1, but not PLD2, is multi-monoubiquitinated. The ubiquitination modification might represent a novel regulatory mechanism in PLD1 functioning, particularly in the context of subcellular trafficking between different membrane compartments.

Phospholipase D (PLD) 2 catalyzes the hydrolysis of phosphatidylcholine to generate phosphatidic acid (PA) and choline (1). PA may directly affect membrane curvature and in doing so facilitate fusion and fission of membrane vesicles (1), or it may act as a lipid second messenger, binding and activating varied signaling proteins, including Raf-1 (2), mTOR (3), and phosphatidylinositol-4-phosphate kinase (4), and promote mitogenic, anti-apoptotic, and chemotactic signals. Two PLD isoforms, PLD1 and PLD2, have been identified in mammalian cells (1,5). Both are membrane-associated, but they exhibit different patterns of subcellular localization and regulate different steps in membrane vesicle trafficking (6). PLD1 co-localizes with perinuclear small vesicles, the Golgi complex, and early endosomes in non-stimulated cells, translocates to plasma membrane (PM) upon stimulation, and then recycles to perinuclear vesicles through endocytosis (7), whereas PLD2 has been reported to localize at the plasma membrane (PM) under resting conditions and translocate to endosomes upon stimulation (8). This difference is believed to contribute to the different cell biological roles undertaken by PLD1 and PLD2. However, the basis for this difference in the localization and function of the PLD isoenzymes remains poorly understood.
Ubiquitination is a key posttranslational modification for many proteins (9,10). Lys 48 -linked polyubiquitination mainly directs proteins to the 26 S proteasome to target them for degradation, but the roles of other ubiquitin linkages, especially monoubiquitination and Lys 63 -linked polyubiquitination, have been well documented to regulate the intracellular trafficking of a broad spectrum of proteins (9,10). Of note, the monoubiquitination of PM proteins, such as tyrosine kinase receptors and G protein-coupled receptors, facilitates their endocytosis, endosomal sorting, and lysosomal degradation (11). Therefore, we became interested in whether ubiquitination is also involved in the regulation of PLD isoenzymes. Accordingly, in preliminary studies, we noted that a fraction of PLD1 protein exhibited retarded migration when examined by Western blotting analysis, suggesting a possibility that PLD1 was ubiquitinated, which we subsequently examined. In this report, we present evidence that PLD1, but not PLD2, is multi-monoubiquitinated and investigate the role of lipase activity and palmitoylation, another posttranslational modification of PLD isoenzymes, in the regulation of PLD1 ubiquitination.

EXPERIMENTAL PROCEDURES
Chemicals, Antibodies, and Reagents-All chemicals and antibodies except those mentioned below were obtained from Sigma. The pharmacological inhibitors of PLD compound 5WO (VU0155056) and 809 (VU0155069) were generous gifts from Dr. H. Alex Brown (Vanderbilt University School of Medicine, Nashville, TN). Antibody against HA tag (3F10) was pur-chased from Roche Applied Science, antibody against Myc tag was from Cell Signaling Technology (Boston, MA), antibody against GM130 was from BD Biosciences, antibodies against PLD1 (H160) and ubiquitin (P4D1) were from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA), and antibody against early endosomal antigen 1 (EEA1) was a generous gift from Dr. Marvin J. Fritzler (University of Calgary). Ham's F-12 medium, Dulbecco's modified Eagle's medium, fetal bovine serum (FBS), tetracycline-reduced FBS, penicillin, and streptomycin were purchased from Invitrogen.
Myc-tagged ubiquitin and the empty vector were generous gifts from Dr. Daniel J. Finley (Harvard Medical School, Boston, MA). FLAG-tagged ubiquitin, Lys 48 mutant, Lys 63 mutant, and lysineless mutant were obtained from the University of Dundee (Scotland, UK). All PLD or PLD mutant vectors were constructed as previously described (7,12). Cells were transiently transfected with these expression vectors using Lipofectamine 2000 as per the manufacturer's specifications (Invitrogen). Protein expression was monitored using Western blotting analysis.
The cDNAs encoding HA-tagged PLD or PLD mutants were inserted into a tetracycline-inducible T-REx system (Invitrogen) in CHO cells (8,13). Expression of PLD was induced in medium containing 1 g/ml doxycycline (Dox). To avoid unintentional induction of PLD expression, the PLD T-REx CHO cells were cultured with tetracycline-reduced FBS.
Immunoprecipitation and Western Blotting Analysis of Ubiquitination of PLD-For immunoprecipitation-ubiquitination experiments, cells were lysed for protein extraction using radioimmune precipitation assay buffer (Sigma) supplemented with 1% (v/v) Triton X-100, protease inhibitor mixtures (Roche Applied Science), 1 mM NaF, 1 mM Na 3 VO 4 , 1 mM phenylmethanesulfonyl fluoride, and 5 M MG132. 1.5 M MG132, an inhibitor of 26 S proteasome, was added for 6 h before cell lysis to prevent the potential degradation of ubiquitinated proteins. An equal amount of protein from each sample was diluted to 1 g/ml with radioimmune precipitation assay buffer and precleaned with protein G-Sepharose beads (GE Healthcare) at 4°C for 30 min (mouse IgG was also added if anti-FLAG tag antibody was used for immunoprecipitation). To immunoprecipitate overexpressed HA-tagged PLD and ubiquitinated species of PLD1, precleaned lysates were subjected to an overnight incubation with antibodies against HA tag and FLAG tag, respectively, and then incubated for 3 h with protein G-Sepharose beads at 4°C. Endogenous PLD1 was immunoprecipitated using overnight incubation with immobilized PLD1 antibody conjugated to protein G-Sepharose beads. Immunoprecipitated proteins were analyzed using SDS-PAGE in gels containing 8 M urea and then transferred to nitrocellulose membranes (Bio-Rad). The membranes were blocked with 5% (w/v) skim milk for 1 h at room temperature, followed by incubation with primary antibodies against Myc tag, FLAG tag, and ubiquitin to detect Myc-tagged, FLAG-tagged, and endogenous ubiquitin, respectively, using a horseradish peroxidase-coupled secondary antibody and chemiluminescence (ECL reagent, GE Healthcare). To ensure that the change of ubiquitination levels of PLD was due to the alteration of levels of ubiquitin, expression levels of ubiquitin were monitored in whole cell lysates using Western blotting analysis. For Western blots without immunoprecipitation, non-reducing sample buffer (50 mM Tris-HCl, pH 6.8, 2% (w/v) SDS, 10% (v/v) glycerol) supplemented with protease inhibitor mixtures (Roche Applied Science), 1 mM NaF, 1 mM Na 3 VO 4 , 1 mM phenylmethanesulfonyl fluoride, and 5 M MG132 was used to extract total cellular protein.
PLD Activity Assay-A PLD activity assay using 1-butanol was conducted as described previously with minor modifications (12). In brief, serum-starved cells were labeled with l Ci/ml [9,10-3 H]myristic acid (GE Healthcare) for 24 h and then washed and pre-equilibrated for 1 h. Phorbol 12-myristate 13-acetate (PMA) (100 nM) and 1-butanol (0.5%, v/v) were added during the last 10 min. Lipid was extracted after incubation as specified in the figure legends, and phosphatidylbutanol was separated on a silica thin layer chromatography sheet. The PLD activity was calculated as the radioactivity counts of phosphatidylbutanol normalized to the counts of total lipids as measured using a liquid scintillation counter (Beckman).
Construction of PLD1 Mutants-Mutagenesis of PLD1 K898R, ⌬N (lacking N-terminal 328 amino acid residues), and C240S,C241S mutants were constructed as previously described (7). Mutagenesis of PLD1 H896R,D903A, ⌬PH (lacking PH domain), and R691G,R695G mutants was performed using the QuikChange kit (Stratagene, La Jolla, CA). A plasmid expressing HA-tagged PLD1 fused N-terminally to Myctagged ubiquitin (Ub-PLD1) was also constructed using the QuikChange kit. Briefly, the Myc-tagged ubiquitin fragment was amplified with Myc-tagged ubiquitin expression vector as template using conventional PCR with the primer 5Ј-CGG GTA TGG CTT CTA GCG CTG GTG AAC AAA AGC TTA TTT C-3Ј (forward) and 5Ј-GGC ACG TCA TAA GGA TAT CTT AGT CTT AAG ACA AGA TG-3Ј (reverse) and then subjected to agarose gel purification using the GeneJET TM gel extraction kit (Fermentas, Burlington, Canada) and inserted into the HA-tagged pCGN-PLD1 plasmid between the promoter and the HA-tagged PLD1 cDNA according to the manufacturer's specifications. Note that two C-terminal glycine residues of ubiquitin were removed to avoid the potential cleavage by deubiquitinating enzymes. All intended mutations were confirmed by sequencing (University of Calgary Core RNA Services).
Immunofluorescence Analysis of PLD1 Localization-COS-7 cells grown on glass coverslips were briefly washed with PBS, fixed in 4% paraformaldehyde for 15 min, permeabilized in 0.2% Triton X-100 for 5 min, and blocked in 2% skim milk for 30 min. Cells were then stained with GM130 or EEA1 antibody for 1 h, followed by incubation with rhodamine-conjugated antimouse secondary antibody (Invitrogen) for 1 h to label cis-Golgi complex and early endosome, respectively. HA-tagged PLD1 was stained with HA tag antibody and Alexa Fluor 488-conju-gated anti-rat secondary antibody (Invitrogen). In some experiments, Myc-tagged ubiquitin-PLD1 fusion protein was stained with Myc tag antibody and rhodamine-conjugated anti-mouse secondary antibody. The coverslips were then mounted in a drop of Prolong antifade reagent containing 4Ј,6-diamidino-2phenylindole (Invitrogen). Imaging and quantification of localization of PLD1 proteins were performed using a ϫ40 dry lens and a Spot Insight Color (3.2.0) CCD digital camera (Diagnostic Instruments, Sterling Heights, MI) with a fluorescence microscope (BX50, Olympus). At least 200 cells were randomly selected and scored for each sample.
Data Analysis-Data are expressed as the mean Ϯ S.D. The number of replicates (n) represents the number of independent experiments performed. Densitometry was conducted using ImageMaster (Amersham Biosciences). Differences between means were evaluated by Student's t test or one-way analysis of variance. p Ͻ 0.05 was considered significant. All statistical analysis was performed using Instat3.0 (GraphPad Software).

RESULTS
Phospholipase D1, but Not Phospholipase D2, Is Ubiquitinated-To examine whether PLD isozymes undergo ubiquitination, we began by employing CHO cells harboring a tetracycline-inducible (T-REx) system to express HA-tagged PLD1 and PLD2 (PLD1/2 T-REx CHO cells) to facilitate our detection of ubiquitination (8,13). Myc-tagged ubiquitin was transfected into PLD1 and PLD2 T-REx cells, followed by induction of PLD expression with doxycycline (Dox; 1 g/ml) for 24 h. MG132 (1.5 M), the inhibitor of 26 S proteasome, was added to prevent the potential rapid degradation of the PLD isozymes, which were immunoprecipitated from whole-cell lysates using anti-HA tag antibody and visualized by Western blotting analysis using anti-HA tag antibody and anti-Myc tag antibody. The expression level of PLD1 was consistently lower than that of PLD2. Upon overexpression of ubiquitin, a broad pattern of ubiquitinated higher molecular weight species was detected in cells induced to express PLD1 but not in cells induced to express PLD2, suggesting that PLD1, but not PLD2, is a target for ubiquitination in vivo ( Fig. 1

, A (top) and B).
We next sought to confirm the PLD1 ubiquitination and determine what percentage of PLD1 became ubiquitinated. To do so, PLD1 T-REx cells were transfected with FLAG-tagged ubiquitin, and the ubiquitin conjugates were immunoprecipitated using anti-FLAG tag antibody. The successful pull-down of ubiquitin conjugates was confirmed by Western blotting analysis using anti-FLAG antibody (Fig. 1C, right). Detection of the higher molecular weight PLD1 species in the immunoprecipitates using HA tag antibody confirmed the ubiquitination of PLD1. Approximately 3.3 Ϯ 0.8% the total PLD1 protein induced by Dox using the T-REx system was ubiquitinated (n ϭ 3).
We then examined the modification of endogenous PLD1 by endogenous ubiquitin using a rabbit polyclonal PLD1 antibody (H-160), which was confirmed to be capable of detecting PLD1 as well as its ubiquitinated species, and immunoprecipitated PLD1 protein (data not shown). PLD1 was immunoprecipitated from HeLa cell protein extracts using the anti-PLD1 antibody (and rabbit IgG as a specificity control), followed by Western blotting analysis with anti-ubiquitin antibody. A broad band of ubiquitin-labeled proteins was observed (Fig. 1D), which was similar in size to that seen when PLD1 and ubiquitin were overexpressed (Fig. 1A), suggesting that endogenous PLD1 also undergoes ubiquitination.
PLD1 Undergoes Multi-monoubiquitination-The type of ubiquitination that takes place on a given protein can result in distinct outcomes (9 -11). We thus examined whether the ubiquitination of PLD1 was through Lys 48 -or Lys 63 -linked ubiquitination or monoubiquitination, by transfecting the PLD1 T-REx CHO cells with FLAG-tagged wild-type ubiquitin, a ubiquitin Lys 48 mutant (K48R), a Lys 63 mutant (K63R), or a lysineless mutant (K0). The induced PLD1 was immunoprecipitated with anti-HA tag antibody, and the ubiquitination was detected with anti-FLAG tag antibody. Transfection with K48R or K63R did not reduce the extent of PLD1 ubiquitination signals ( Fig. 2A), indicating that PLD1 does not undergo Lys 48 -or Lys 63 -linked ubiquitination. Moreover, the ubiquitination signals were also not reduced in cells expressing the K0 mutant ( Fig. 2B), indicating that PLD1 undergoes only monoubiquitination, presumably multiple times on individual PLD1 lysine residues.
PLD Lipase Activity Is Critical for the Ubiquitination of PLD1-Multi-monoubiquitination frequently correlates with activity of the target proteins, such as ligand-induced activation of tyrosine kinase receptors (11,14). We thus examined whether the ability of PLD1 to perform its enzymatic lipase activity affected the extent of its monoubiquitination. To do so, we first examined the effects of inhibition of PLD activity on PLD1 multimonoubiquitination using two recently developed pharmacological inhibitors (15), compound 5WO (VU0155056, 10 M, dual inhibitor of PLD1 and PLD2) and compound 809 (VU0155069, 10 M, PLD1-specific). Our results showed that inhibition of PLD1 and PLD2 or PLD1 alone significantly reduced the ubiquitination of PLD1 (Fig. 3, A and B).
To further establish the requirement of PLD1 lipase activity for its ubiquitination, we took advantage of the lipase-inactive HKD motif mutant, PLD1 K898R (16). The mutant was stably expressed in CHO cells using the T-REx inducible system, as described earlier for PLD1 and PLD2 and shown in Fig. 1A. After transfected with Myc-tagged ubiquitin, the cells were induced by the addition of Dox. Thereafter, ubiquitination of the mutant PLD1 was detected using immunoprecipitation and Western blotting analysis as described before. Supporting the results obtained using pharmacological inhibitors, we observed significant reduction of the ubiquitination signal in PLD1 K898R-expressing cells as compared with cells expressing wildtype PLD1 (Fig. 1, A and B). Because lysine 898 could be a direct ubiquitin conjugation site, we generated several additional non-lysine lipase-inactive mutants to confirm the activity dependence of PLD1 ubiquitination, including two HKD motif mutants, H896R and D903A (16), and a phosphatidylinositol 4,5-bisphosphate non-interacting mutant, PLD1 R691G,R695G (7,17). All three inactive mutants showed significant reduction in ubiquitination signals when transiently expressed in CHO cells together with Myc-tagged ubiquitin (Fig. 3, C-E), indicating that the lipase activity of PLD1 is critical for its monoubiquitination.
Because catalytic activity was important, but the basal activity of PLD1 is low (1), we employed PMA to stimulate PLD1 activity and determined if PMA increased the extent of ubiquitination. However, 100 nM PMA treatment, although enhancing PLD activity (Fig. 4C), did not stimulate the ubiquitination of either HA-tagged PLD1 in CHO cells (Fig. 4, A and B) or endogenous PLD1 in HeLa cells (Fig. 4D).
Palmitoylation Is Critical for the Ubiquitination of PLD1-Palmitoylation of PLD1 is required for its correct membrane localization and endocytosis after PMA-stimulated PM translocation (7,18). Both palmitoylation and monoubiquitination play important roles in the vesicular trafficking of membraneassociated proteins, and palmitoylation has been reported as a prerequisite for the ubiquitination and trafficking of H-Ras before cell lysis to prevent the potential degradation of ubiquitinated PLD proteins. Whole cell lysates were immunoprecipitated using HA tag antibody. The immunoprecipitates were subjected to SDS-PAGE containing 8 M urea, followed by detection with Myc tag antibody for PLD ubiquitination and HA tag antibody for PLD expression. An equal expression level of ubiquitin was monitored by detection of ubiquitin expression in whole cell lysates using Western blotting. IP, immunoprecipitation; IB, immunoblotting. The data are representative of three independent experiments. B, the densitograph of PLD ubiquitination levels was generated from normalization of Myc-ubiquitin signal by HA-PLD signal, as shown in A, from three independent experiments. The ubiquitination level in the wild-type PLD1 group was set as 100%. *, p Ͻ 0.01 when compared with PLD1 ubiquitination level (n ϭ 3). C, representative Western blots showing the ubiquitin-conjugated species of PLD1. PLD1 T-REx CHO cells were transfected with FLAG-tagged ubiquitin, followed by induction by 1 g/ml Dox for 24 h. MG132 (1.5 M) was added for 6 h before cell lysis. The immunoprecipitates obtained with anti-FLAG tag antibody were subjected to urea-SDS-PAGE and immunoblotting with HA tag antibody to detect ubiquitin-conjugated PLD and FLAG tag antibody to detect immunoprecipitated ubiquitin conjugates. 3% total input protein was used to confirm the position and estimate the amount of PLD1 protein ubiquitinated. The data are representative of three independent experiments. D, representative Western blots showing the ubiquitination of endogenous PLD1. HeLa cells were treated with 1.5 M MG132 for 6 h before cell lysis. Whole cell lysates were then immunoprecipitated using rabbit polyclonal PLD1 antibody or rabbit IgG. The immunoprecipitates were subjected to urea-SDS-PAGE, and the ubiquitination and expression of PLD1 were detected with ubiquitin and PLD1 antibodies, respectively.   19). To test the possibility that loss of palmitoylation reduces the monoubiquitination of PLD1, we transfected CHO cells with Myc-tagged ubiquitin and HA-tagged wild-type PLD1 or three PLD1 mutants, PLD1 lacking the first 328 amino acids at the NH 2 terminus (PLD1⌬N), PLD1 lacking the PH domain (PLD1⌬PH), and the palmitoylation-deficient PLD1 C240S,C241S (18,20), all three of which lack palmitoylation and exhibit defects in poststimulation endocytosis (7). Dramatic reduction of ubiquitination was observed for all of the mutants (Fig. 5), indicating that palmitoylation of PLD1 is critical for its monoubiquitination and suggesting an association of monoubiquitination with vesicular trafficking of PLD1. In addition, we observed that the PLD1⌬N mutant showed a nearly complete loss of ubiquitination, whereas PLD1 ⌬PH and C240S,C241S mutants had notable ubiquitination signals, suggesting that some other region or domain modifications in the N terminus may be necessary for ubiquitination.

Ubiquitinated PLD1 Induces Aberrantly Enlarged Vesicles in COS-7
Cells-To determine whether the monoubiquitination of PLD1 plays any role in its subcellular localization, we constructed an N-terminally Myc-tagged ubiquitin-fused HA-tagged PLD1 expression plasmid and then examined the subcellular localization of ubiquitinated PLD1 (Ub-PLD1) in COS-7 cells. Consistent with a previous report (7), ϳ90% of cells transfected with wild-type PLD1 under basal conditions exhibited a perinuclear vesicular distribution for PLD1, as demonstrated by HA tag staining (Fig. 6, A-C and G-I), which partially colocalized with the Golgi complex and early endosomes, as indicated by overlap with GM130 (Fig. 6, A-C) and EEA1 (Fig. 6, G-I). Notably, ϳ50% cells transfected with Ub-PLD1 exhibited an aberrantly enlarged vesicular distribution for PLD1 as demonstrated by HA tag staining (Fig. 6, D-F and J-L) or Myc tag (Fig. 6O). In most Ub-PLD1-positive cells, Ub-PLD1 remained partially co-localized with Golgi complex (Fig. 6, D-F) but generally lost the co-localization with early endosomes (Fig. 6, J-L). To validate the protein expression of Ub-PLD1 in COS-7 cells, we found that Ub-PLD1 co-migrated with the upper band of PLD1 in urea-SDSpolyacrylamide gel when detected with HA tag antibody in Western blotting analysis (Fig. 6N), suggesting that the ectopic ubiquitin conjugation to the N terminus of PLD1 did not lead to its unscheduled polyubiquitination and proteasomal degradation. Additionally, when detected with Myc tag antibody in Western blotting assay, Ub-PLD1, but not wild-type PLD1, showed a positive band of similar size to the HA tag band (Fig. 6N), suggesting that the Myc-tagged ubiquitin fusion was not cleaved by deubiquitinating enzymes.
Because PA is involved in SNARE-mediated membrane fusion (21), we hypothesized that these aberrantly enlarged vesicles might be induced by accumulated PA in situ produced by ubiquitinated PLD1 trapped in some vesicular compartment. To test this possibility, we constructed an N-terminally ubiquitin-fused PLD1 K898R (Ub-PLD1 K898R) expression vector and confirmed that its protein expression and Myc-tagged ubiquitin conjugation are indistinguishable from that of Ub-PLD1 (data not shown). We observed that only 22 Ϯ 12% (n ϭ 3) cells transfected with Ub-PLD1 K898R showed aberrantly enlarged vesicles, and this percentage was significantly lower  Fig. 1. E, densitograph of PLD1 ubiquitination levels in C and D from three independent experiments. Ubiquitination level in wild-type group was set as 100%. *, p Ͻ 0.05 when compared with wild-type group (n ϭ 3). IP, immunoprecipitation; IB, immunoblotting; wt, wild type. than that of cells transfected with Ub-PLD1 (52 Ϯ 5%, n ϭ 3) (Fig. 6M). In addition, 4-h treatment with PLD inhibitor compound 809 (VU0155069; 10 M) and 1-butanol (1%, v/v) decreased the percentage of vesicles containing Ub-PLD1-positive cells (23% for compound 809 versus 59% for DMSO vehicle; 27% for 1-butanol versus 47% for 2-butanol). Taken together, our data indicate that the aberrantly enlarged vesicles in Ub-PLD1-transfected cells are at least partially dependent on PLD1-generated PA, and they may represent some specific vesicular compartments enriched in ubiquitinated species of PLD1.
PLD1 Is Degraded through the Proteasomal Pathway but Not via Lysosomes-Monoubiquitination can serve as a sorting signal to lysosomal degradation for internalized proteins (11). To determine whether the monoubiquitination of PLD1 directs its degradation, we examined the effects of various protease inhibitors on the expression levels of PLD isoenzymes using the Doxinducible system. In serum-starved cells, expression levels of PLD1 were significantly increased at 6 and 24 h (Fig. 7, A and D Fig. 1. B, the densitograph of PLD1 ubiquitination levels in A from three independent experiments. The ubiquitination level in the wild-type PLD1 group was set as 100%. *, p Ͻ 0.01 when compared with wild-type group (n ϭ 3). IP, immunoprecipitation; IB, immunoblotting. APRIL 30, 2010 • VOLUME 285 • NUMBER 18 g/ml)), calpain (PD150606, 50 M), cysteine proteases (E-64, 100 M) or caspase (caspase inhibitor I, 10 M), suggesting that PLD1 is degraded exclusively via proteasomes. Importantly, significant increases in the expression levels of PLD1 K898R and PLD2 were not detected in the presence of the proteasome inhibitors (Fig. 7, B-D).

PLD1 Monoubiquitination
To confirm the proteasomal degradation of endogenous PLD1, serum-starved HeLa cells were treated with cycloheximide (10 M; inhibitor of protein synthesis). Degradation of PLD1 protein became evident by 48 h and substantial by 96 h. The degradation was blocked by MG132 (0.1-2 M) in a dosedependent manner (Fig. 8) but not by chloroquine (100 M) or E-64 (100 M) (data not shown).

DISCUSSION
Monoubiquitination is a critical regulatory mechanism in a variety of cellular processes, including endocytosis, lysosome sorting, transcription, and DNA repair (11). Our results reveal that PLD1, but not PLD2, undergoes multi-monoubiquitination, which may contribute to the differential trafficking behaviors and the roles in cellular signaling of both PLD isoenzymes.
The ubiquitination of PLD1 was confirmed through immunoprecipitation of PLD1 and immunoblotting with anti-ubiquitin antibody. We estimate that the amount of ubiquitinated PLD1 was about 3% of total PLD1 overexpressed in the CHO cells. Of note, this percentage of ubiquitinated species in the total protein population is not unusually low among those proteins undergoing vesicular trafficking facilitated by ubiquitination, such as H-Ras (ϳ2% in CHOK1 cells; the percentage is estimated based on densitometric analysis of published immunoblots) (19), epsin (ϳ6% in 293T cells) (22), Gap1p (ϳ7% in yeast) (23), and Ste2p (ϳ8% in yeast) (24).
An intriguing finding is the role of lipase activity in the regulation of monoubiquitination of PLD1. The dependence on basal lipase activity of PLD1 ubiquitination was validated by the inhibitory effects on PLD1 ubiquitination of the newly developed PLD inhibitors and mutation of the PLD1 HKD motif (His 986 , Lys 898 , and Asp 903 ) (16) or phosphatidylinositol 4,5-bisphosphate binding motif (Arg 691 and Arg 695 ) (7). On the other hand, the inability of PMA to enhance the ubiquitination of PLD1 suggested that enhanced activation by stimuli might not be required for ubiquitination. It is also possible that the ubiquitination has reached the maximal level even under resting conditions. PLD lipase activity normally results in the production of PA. However, the enzyme can also use primary alcohols, such as ethanol or 1-butanol, and when concentrations of the alcohols are sufficiently high, PA production is diminished in preference for generating the corresponding phosphatidylalcohol. Thus, although PLD1 enzymatic activity per se is unaltered, PA production is blocked. Interestingly, culture of CHO cells in 0.5% 1-butanol did not impair PLD1 ubiquitination, 3 supporting the speculation that the active conformation of PLD1 and/or access to its catalytic site, but not production of PA, is critical for its ubiquitination. It is unknown 3 H. Yin and X.-L. Zheng, unpublished observation. whether PLD1 undergoes conformational changes in the presence of its substrate, phosphatidylcholine, or upon stimulation. However, it is possible that the PLD1 protein requires a specific active conformation to be recognized by the ubiquitination machinery, which would be consistent with our findings. We also found that PLD2 does not become ubiquitinated. We thus examined whether the loop domain, one of the PLD1-unique regions, is the domain that is targeted by or responsible for the ubiquitination but found that PLD1 lacking the loop domain only showed a modest reduction (25 Ϯ 11%, n ϭ 3) in ubiquitination in comparison with wild-type PLD1, 3 indicating that other regions of the PLD1 protein are responsible for the difference in ubiquitination between the PLD isoenzymes.
Both palmitoylation and monoubiquitination are important regulatory mechanisms in the intracellular trafficking of target proteins (11,25). In our study, the monoubiquitination of PLD1 was found to be dependent on its palmitoylation. A previous study indicates that loss of palmitoylation abolishes the lipid raft association and poststimulation endocytosis of PLD1 (7,18). This suggests that monoubiquitination of PLD1 may play an important role in its intracellular trafficking. Our data on PLD1 have echoed findings with H-Ras that palmitoylation is required for its mono-and diubiquitination and that the cooperation of palmitoylation and ubiquitination modulates its trafficking and signaling (19). It remains unclear how palmitoylation promotes monoubiquitination of PLD1. Because the palmitoylation-deficient PLD1 mutant is still lipase-active (7), it is likely that palmitoylation regulates the monoubiquitination through altering the membrane localization of PLD1 (e.g. its lipid raft association (7)) to facilitate contact of PLD1 with the ubiquitination machinery or isolate PLD1 from deubiquitinating enzymes.
The biological relevance of PLD1 ubiquitination has been supported by the association of ubiquitinated PLD1 with an aberrantly enlarged vesicular compartment. Interestingly, this observation is reminiscent of membrane trafficking defects resulting from malfunction of ARF1 (26), a PLD1 activator, or endosomal sorting complex required for transport machinery, such as SKD1/VPS4 (27) and CHMP2B (28). The enlarged vesicle observed in our study is possibly post-Golgi but not early endosome, as demonstrated by a lack of overlapping with EEA1 staining. It remains to be determined whether these vesicles originate from endoplasmic reticulum, late endosome, multivesicular body, or lysosome. The reduced formation of enlarged vesicles upon Ub-PLD1 K898R expression or blockade of PA production by PLD inhibitor or 1-butanol suggests that in situ production of PA contributes to the enlargement of these vesicles, possibly through SNARE-dependent membrane fusion events (21). In addition, non-lipase PLD1 functions, such as interaction with lipids or proteins and recently identified dynamin GAP activity (29), could also contribute to this vesicle phenotype because K898R mutation showed an incomplete suppression effect on the vesicle enlargement. It is possible that the enlarged vesicle observed in the current study may represent an intermediate vesicular compartment enriched in ubiquitinated species of PLD1 during PLD1 trafficking. Since the ectopically ubiquitinated PLD1 mutant employed in our study is resistant to deubiquitination, our data suggest that an ubiquitination/deubiquitination cycle may regulate the trafficking of PLD1 at different membrane fractions. Of note, PLD1 plays a regulatory role in the vesicular trafficking of various membrane-associated proteins, such as endocytosis of EGFR (30), PM targeting of Rap1 (31) and GLUT4 (32), and exocytosis of PLD expression level was detected using HA tag antibody. ␣-Tubulin was used as loading control. The data are representative of three independent experiments. D, the densitographs of PLD expression were generated from normalization of HA-PLD signal by the ␣-tubulin signal as shown in A-C from five independent experiments. The expression level of PLD isoenzyme in the control group of each panel was set as 100%. *, p Ͻ 0.05 when compared with control group (n ϭ 5). FIGURE 8. Endogenous PLD1 is degraded by proteasomes. Representative Western blots show the degradation of endogenous PLD1 and the effects of MG132 (0.1-2 M). Serum-starved HeLa cells were treated with cycloheximide (CHX; 10 M) for different time periods in the presence of MG132 as indicated before cell lysis. Culture medium was changed every day. Total cell extracts were subjected to urea-SDS-PAGE. PLD1 protein level was detected using PLD1 antibody. ␤-Actin was used as a loading control. The remaining PLD1 level was calculated as PLD1 signal normalized by ␤-actin. The PLD1 level for each inhibitor treatment at 0 h was set as 100%. The data are representative of three independent experiments. noradrenalin (33). It will be interesting to assess whether the multi-monoubiquitinated species of PLD1 represents the subset of the PLD1 protein actively engaged in membrane trafficking. Ubiquitinated PLD1 could be localized in some specific membrane fractions and act as a membrane-processing component through local PA production. Consistent with this proposal, PA is enriched in some specific post-Golgi vesicles (34).
Monoubiquitination is widely recognized as a signal for the lysosomal sorting and degradation (11) and is also now known to play a role in proteasomal degradation, as has been reported for ferritin (35). In our study, PLD1 was degraded via proteasomes rather than lysosomes, whereas in contrast, PLD1 K898R and PLD2, both of which were defective in ubiquitination, did not undergo proteasomal degradation, suggesting that the proteasome degradation is dependent on PLD1 monoubiquitination. Because elevated PLD1 activity has been reported in many human diseases (36), it will be interesting to evaluate the pathophysiological roles of the ubiquitin-dependent degradation of PLD1.
The E3 ligase and deubiquitinating enzyme for PLD1 are unknown. Literature analysis suggests that the Nedd4 family might be candidate of E3 ligase for PLD1. Nedd4 may monoubiquitinate PLD1 at the Golgi complex so as to regulate its endosomal sorting, as it does for the Gap1p amino acid permease (23). UBPY (37) or AMSH (associated molecule with the SH3 domain of STAM) (38) could be a deubiquitinating enzyme for PLD1 because both can deubiquitinate ubiquitinated cargoes during endosomal sorting.